[ENBD] 20G dd hang up!

[Peter T. Breuer]

"A month of sundays ago Kuniyasu SUZAKI wrote:"
> And I could read full volume from client from server and it was succeeded.
>     dd if=/dev/nda of=/dev/null bs=1048576 count=19539
> 
> The option of nbd-server and nbd-client is as follows.
>   On server machine:
>     nbd-server 5058 /dev/sda1  -t 120 -b 4096  -0
>   On client machine:
>     nbd-client london:5058  -n 8 -b 4096  -t 120  -p 1  -d 1 /dev/nda

Try with -b 1024. Some people seem to have problems with 4096
blksize. I have no idea why (it cannot _in principle_ be internal to
enbd).

> Can I fix the "dd" write problem to tune time out option (-t) or something?
> Please give me suggestion or idea to fix the problem.

Well, there are things you can do here to turn the problem into one
that is more easily analized, but see first if it goes away with 1K
blocksize.

BTW, you should be using  enbd 1.4.16 or 2.4.17pre (not 2.4.6). Also
try that (first ;-).

> The machines are DELL workstations with Pentium III 933Mhz with 512M
> memory and 100M Ether NIC(3Com 3c905C Tornado). I connect the two
> machines with cross ether cable. The Network performance which was
> measured by netperf was 95 Mbps. It was nice environment, I think.

> ------
>   Kuniyasu SUZAKI, National Institute of Advanced Industrial Science and Technology,
>   Tsukuba Central 2, Umezono 1-1-1, Tsukuba, Ibaraki 305-8568, JAPAN
>   Project NTC (Network Transferable Computer) http://www.etl.go.jp/~suzaki/English/NTC

Peter